(Press-News.org) A machine learning-based weather forecasting model from Google DeepMind leads to better, faster, and more accessible 10-day weather predictions than existing approaches, according to a new study. The model, dubbed "GraphCast," outperformed traditional systems in 90% of tested cases. It also performed well in predictions related to extreme events, for which it was not directly trained. "We believe this marks a turning point in weather forecasting," write the authors. The gold-standard approach for weather forecasting today is "numerical weather prediction (NWP)." While the accuracy of NWP models has improved drastically over the several decades, they are costly, complex, and computationally demanding, requiring increased compute resources to improve forecast accuracy. Machine learning-based weather prediction (MLWP), which can be trained from historical data, offers an alternative. Moreover, MLWP can leverage modern deep-learning hardware for efficiency.
Here, DeepMind researchers led by Remi Lam introduce GraphCast, a machine learning-based method for medium-range weather prediction trained directly from reanalysis data of past atmospheric conditions. Implemented as a neural network, it can produce an accurate 10-day forecast in under a minute on a single TPU chip. As input, GraphCast takes the two most recent states of Earth’s weather – the current time and six hours earlier – and predicts the next state of the weather six hours ahead, providing global weather prediction coverage at a roughly 0.25° latitude/longitude resolution. These predictions can be fed back into the model as inputs to generate a longer trajectory of weather states. To evaluate the forecast skill of GraphCast, Lam et al. compared GraphCast’s accuracy to that of HRES – the most accurate deterministic medium-range weather forecasting model currently used – on a large number of weather variables and lead times. They found that GraphCast was able to significantly outperform HRES on 90% of 1380 verification targets. What’s more, the platform was better at predicting severe events, including tropical cyclone tracks, atmospheric rivers (narrow regions of the atmosphere responsible for significant poleward water vapor transport), and extreme temperature anomalies, despite not being specifically trained on them. "Our approach should not be regarded as a replacement for traditional weather forecasting methods," say the authors. "Rather our work should be interpreted as evidence that MLWP is able to meet the challenges of real-world forecasting problems and has potential to complement and improve the current best methods."
END
Introducing: AI-powered medium-range weather forecasting from Google DeepMind
2023-11-14
ELSE PRESS RELEASES FROM THIS DATE:
Brain CareNotes telehealth app supports dementia caregivers
2023-11-14
INDIANAPOLIS — With the number of Americans living with dementia expected to more than double to 13.8 million by 2060, two Regenstrief Institute research scientists and Indiana University professors have created an app to ease the burden on caregivers.
In 2021, more than 11 million family members or other unpaid, informal caregivers provided nearly 16 billion hours of care to people with dementia. Globally, approximately 50 million people are affected by this neurocognitive disorder.
Richard Holden, PhD, M.S., and Malaz Boustani, M.D., MPH, have created and real-world tested the evidence-based Brain CareNotes. This easy-to-use app will ...
Korea University and Ewha Womans University researchers highlight advancements in biomedical research with enzyme-activated fluorescent probes
2023-11-14
Enzymes, essential for normal cellular and physiological functions, are implicated in various diseases like cancer and diabetes due to their abnormal activity. Therefore, tracking enzyme activity is a valuable strategy for the diagnosis and monitoring of diseases. Conventional imaging techniques are limited by the need for contrast agents, low sensitivity, and spatio-temporal resolution. To overcome these limitations, researchers are increasingly investigating fluorescent probes for non-invasive and real-time visualization of enzyme dynamics and corresponding disease status.
In a new review article, researchers from Korea have summarized the latest advancements ...
A “gold standard” for computational materials science codes
2023-11-14
For the past few decades, physicists and materials scientists around the world have been busy developing computer codes that simulate the key properties of materials, and they can now choose from a whole family of such tools, using them to publish tens of thousands of scientific articles per year. These codes are typically based on density-functional theory (DFT), a modelling method that uses several approximations to reduce the otherwise mind-boggling complexity of calculating the behavior of each individual electron according to the laws of quantum mechanics. The ...
Taylor & Francis partnership with the National Council on Measurement in Education yields dramatic open access results
2023-11-14
Collaboration between the National Council on Measurement in Education (NCME) and Taylor & Francis (T&F) to make more books in Education open access catapulted readership six-fold in the partnership’s first year.
The NCME Applications of Educational Measurement and Assessment series, available here, provides definitive research, theory, and applied insights in educational measurement, assessment, testing, and psychometrics. Topics addressed include validation, fairness, accountability, technology, natural language processing, and beyond.
Ten edited volumes have been published since the partnership between NCME and T&F commenced in 2011, and the ...
Kissick Family Foundation, Milken Institute announce $2.5 million in funding for dementia research
2023-11-14
WASHINGTON, DC (November 14, 2023)—The Kissick Family Foundation Frontotemporal Dementia (FTD) Grant Program, in partnership with the Milken Institute, launched its first-ever request for funding proposals today. Up to $2.5 million in total funding will be made available to researchers from around the world whose work aims to increase scientific understanding of FTD. The program is accepting applications for two-year research projects and intends to award three to five grants to doctorate-level ...
Study reduces ‘vivid imagery’ that fuels addiction cravings
2023-11-14
In 2021, 61.2 million Americans ages 12 and older used illicit drugs and more than 106,000 died from drug-involved overuse. Florida ranks second to California, with 5,300 annual overdose deaths, outpacing the national average by about 23 percent. Excessive alcohol use is the leading cause of preventable death in the United States and is associated with numerous health issues, including heart disease, cancer and poor mental health.
Although individuals with substance use disorder use outpatient sober support services and resources, relapse rates are still as high as 70 percent, indicating a need for more treatment modalities.
A new Florida Atlantic ...
$3M to boost state-of-the-art solar manufacturing
2023-11-14
Images
A new breed of semiconductors that could enable breakthroughs in solar cells and LEDs will benefit from cutting-edge manufacturing approaches, through a new project led by the University of Michigan.
Backed by $3 million from the National Science Foundation, it includes partners at the University of California San Diego.
The effort combines hands-on work that improves upon the process of layer-by-layer deposition of semiconductor materials during production with an information-sharing ...
Mount Sinai bioengineers send cardiac muscle samples into space to study heart cell biology in microgravity
2023-11-14
Mount Sinai’s Cardiovascular Research Institute is sending bioengineered human heart muscle cells and micro-tissues into space for the first time on NASA’s 29th SpaceX commercial resupply services mission, which launched Thursday, November 9. The “SpaceX CRS-29” mission is sending scientific research to the International Space Station (ISS), where the samples will stay for approximately 30 days before returning to Earth.
Through this experiment, Icahn School of Medicine at Mount Sinai researchers aim to gain a better understanding of how cardiac muscle cells, or cardiomyocytes, adapt to extreme biological stresses, and how microgravity and other features of ...
Long-term heat tolerance in plants
2023-11-14
Two papers describe the genetic basis of long-term heat tolerance in the model plant Arabidopsis thaliana, with implications for crop breeding. Teruaki Taji and colleagues evaluated dozens of lines of the model mustard weed for both long term (37 °C for 36 days) and short term (42°C for 50 minutes) heat stress. The authors found considerable variation within the species, but little overlap between responses to the two different heat challenges, suggesting that long-term heat stress tolerance is controlled by different cellular mechanisms than the more commonly studied short-term heat stress. Chromosomal mapping using the F2 progeny of a cross between a long-term-heat ...
Health plays a role in older adults’ vulnerability to scams, poll suggests
2023-11-14
Three out of every four older adults say they have experienced a fraud attempt by phone, text, email, mail or online in the last two years, a new poll shows. Three in ten say they’ve been victims of at least one scam.
The poll reveals an especially strong link between an older adult’s health and their vulnerability to scams – both being able to spot one and becoming the victim of one.
Across the board, people aged 50 to 80 who reported being in fair or poor physical or mental health, those with disabilities, and those who rate their memory as fair or poor were more likely than others their age to say they’d experienced ...